Companies hold immense amounts of unstructured data, including digital documents, data relevant to the documents, and engineering drawings. This makes it very difficult to identify which documents are related to each other, meaning they usually have no choice but to depend on the knowledge and experience of individual experts. For example, when a law is changed, many of our customers in the manufacturing industry need to check which products and specifications are affected by the change by referring to relevant data, which entails a great deal of time. These tasks are dependent on the knowledge and experience of each individual, which results in variations between people in the time required to complete the tasks and the quality of work.
Our original data linking*1 technology allows terms from different texts to be extracted and associated with each other based on their semantic relations by using our original natural language processing technology. These extracted and associated terms are then structured into an ontology.*2 Using this ontology makes it possible to quickly identify relevant documents.
- *1 The linking of terms from different documents based on their semantic relations.
- *2 Data is structured to show how individual terms or concepts are related to each other and to allow those relations to be processed by a computer.
For example, at FUJIFILM Business Innovation, we have succeeded in cutting down the time required for specialists to prepare application forms required by law by 80% by automating the tasks for checking which engineering drawings are related to which component data. Being able to quickly identify relevant documents has made it considerably easier to adapt to changes made to laws. Tasks that were once dependent on specialist knowledge and experience can now be carried out by anyone, leading to improved efficiency of work and risk reduction.